increased shear, reduced wall temperatures, use of …
TRANSCRIPT
Proceedings of International Conference on Heat Eχchanger Fouling and Cleaning - 2013 (Peer-reviewed)
June 09 - 14, 2013, Budapest, Hungary
Editors: M.R. Malayeri, H. Muller-Steinhagen and A.P. Watkinson
Published online
www.heate χchanger-fouling.com
INCREASED SHEAR, REDUCED WALL TEMPERATURES, USE OF hiTRAN WIRE
MATRIX INSERTS IN SYSTEMS SUBJECT TO FOULING
p. Drogemuller^ W. Osley l and D. PhiUpp^
1 Cal Gavin LTD, Minerva Mill Technology Centre, Station Road, Alcester, B980PW, UK
[email protected], [email protected]
^Technische Universitat Braunschweig, Institute for Chemical and Thermal Process Engineering (ICTV)
ABSTRACT
Fouling characteristics are influenced largely by the
properties of the 万thermal and hydrodynamic boundary
layers. This paper demonstrates that tube side heat transfer
enhancement devices such as hiTRAN thermal systems can
be used as a tool to influence both wall temperatures and
wall shear stress. Eχperimental results from Laser Doppler
Velocimetry experiments and direct wall shear
measurements with hiTRAN Wire Matrix inserts are
presented. These measurements were verified by CFD
simulations. The results are comparable in good agreement
and indicate an increase in wall shear forces when applying
this insert type. The main driving force for the use of
hiTRAN Elements in heat exchanger design is the
substantial increase in tube side heat transfer performance.
As a direct consequence of increasing the rate of heat
transfer, the resulting tube wall temperature is changed and
therefore the fouling behaviour. In a real case crude oil
fouling scenario hiTRAN is used in order to improve the
overall duty of the unit with increased wall shear and
reduced wall temperatures.
INTRODUCTION
A body of research has been undertaken to identify the
parameters which determine fouling behaviour. As a result
semi empirical fouling rate models have been developed. To
describe crude oU fouling behaviour over time, the Ebert
and Panchal Model (Ebert and Panchal 1997)and its
variants have been successfully applied. These models are
developed to describe chemical reaction fouling in tube side
flows (Wilson et ai, 2012). Some research also takes into
account the use of tube \internals in order to achieve
threshold conditions (Mengyan et a/, 2012). Common to
these types of model is a deposition and a suppression term.
Since crude oil fouling downstream from the desalter can be
characterized by chemical reaction fouling the reaction rate
of the foulant will increase with higher temperatures. The
deposition term is strongly influenced by the film
temperature of the fluid. The suppression term describes
how the influence of wall shear forces counteracts the
buildup of foulant and supresses further buUd up of fouling
deposit:
笠 ゜αj?eβexp
0
0
0
0
0
0
Lr>
0
in
o
4
3
3
2
2
[
D
o
I
9
Jn
ie
j3
d
ujai
E
-i 150
100
(も卜 、
0
lower Film temperature
10 20 30 40 50
(1)
60
Wall sKear stress【Pal
Fig. 1 : Threshold model and implication of additional
shear stress and lower film (wall)temperatures。
By default this type of model implies that in a technical
process film temperatures and wall shear stress can be
found to suppress fouling entirely as shown in fig. 1. The
fouling model parameters a,」 β,Y as well as the Activation
Energy Ea have to be determined by experiments from
laboratory or process data。
In a plain empty tube exchanger design, the parameters
for wall temperature and wall shear are determined by the
process, geometry and property conditions. hiTRAN tube
inserts can be used in order to beneficially influence these
two parameters.
hiTRAN WIRE MATRIX ELEMENTS
hiTRAN technology (fig. 2)is used effectively to
increase the turbulence of tube side flow. Main applications
are in laminar and transitional flow regimes, but hiTRAN
elements are also used in turbulent flow in cases were
additional pressure drop is available within the system.
Pressure drop and heat transfer characteristics are adapted to
the process requirement by changing the packing density of
the inserts. This insert type has been applied successfully in
fouling
2007).
Drogemij ‖er et al. /Increased shear, reduced wall temperatures, use of
applications (Gough et a/., 1995, Ritchie et a/.
Fig. 2: Typical hiTRAN Wire Matrix element.
INFLUENCE OF HEAT TRANSFER ON WALL
TEMPERATURES
Tube side heat transfer measurements at a single phase
test facility were conducted in order to measure the tube
side performance for a wide range of hiTRAN insert
geometries with varying packing densities.
1000
E
O
T-
u-
Jd
. n
z
1
』o
p
e
i-f
1
皿l
10 100 1O00
high p●kli罵itきitHY
10000 100000
Reynolds Number [・l
Fig. 3: Typical heat transfer performance of two hiTRAN
geometries.
In figure 3 the measured heat transfer results for two
hiTRAN insert geometries with low and high packing
densities as a function of Reynolds are shown. Thermal and
hydraulic performance data for heat transfer and pressure
drop for the full range of hiTRAN geometries are
implemented in the Software hiTRAN.SP distributed by Cal
Gavin LTD. It can be seen that increases of up to 16 times
heat transfer rate in laminar flow and 2 to 4 times the heat
transfer rate in turbulent flow are measured. Neglecting wall
resistance and any fouling layer, the wall temperature can
be eχpressed as follows:
7し -
- 弓K +
づ (石べ) (2)
For an increased tube side heat transfer coefficient the
wall temperature approaches the tube side bulk temperature.
In case the tube side flow is heated a lower wall temperature
is achieved. The increased tube side heat transfer can be
utilized in order to move into the area of no fouling in figure
1
WALL SHEAR STRESS WITH hiTRAN
In plain empty tubes the wall shear stress てW is directly
linked to the pressure drop Ap:
Tw =! 牛旦 (3)
When applying tube inserts a substantial amount of
pressure drop is generated by drag and shear forces at the
insert geometry. These forces have to be known in order to
determine the wall shear stress based on pressure drop
measurements. This paper describes three independent
methods to measure and also simulate the tube side wall
shear stress in the presence of hiTRAN inserts.
Wall shear stress calculatiion by measurement of the
velocity field
Wall shear stress is directly associated with the velocity
field near to the tube wall:
du・ (4 )
U / Umax 【-】
』 ・
f
- n o
ハ/へ :: ハ
ド / X/ v: ダ Xj x:L /', ふ へ y i l
Radial position 【mm]
Fig.. 4: Normalized velocity profile at Reynolds Number
500 just after hiTRAN insert. For plain tube
comparison Yellow line - turbulent profile, pink
line -laminar profile are shown
www.heatexchanger-fouling.com
Heat Exchanger Fouling and Cleaning - 2013
Laser Doppler Velocimetiy (LDV)data from Smeethe
et al (2004)were re-evaluated. He measured the velocity
profile immediately downstream of hiTRAN inserts.
Aluminum tracer particles were added to the flow without
slip and subjected to a dual beam laser. With this technique
the velocity of the tracer particles and therefore the flow
profile was measured. The velocity profile indicates a much
steeper velocity gradient du/dy at the wall compared to a
laminar profile at the measured Reynolds number (fig. 4). In
fact at a Reynolds number 500 the velocity gradient near to
the wall with hiTRAN was similar to a profile expected for
turbulent flow conditions.
Near to the wall detailed measurements were carried
out at different Reynolds numbers for the plain empty tube
and medium dense insert geometry. The results are shown
in figure 5
1
0
8
0
6
0
4
I
E
『
一
/n
0.2
0
今-Re3000
・‐-
.^
-
Re500
RelOO
laminar
theory
0 0,2 0.4 0.6 0.8 1
Wall distance fmm]
Fig. 5.: Near wall velocity profile for different Reynolds
numbers compared to plain empty tube theory for a
constant insert geometry
For the plain empty tube a normalized velocity gradient
du/dy of ~0.2 1/s is measured and calculated in laminar
flow. For transitional flow at Reynolds 3000 this value is
about 0.275 1/s. With hiTRAN this gradient is steeper,
generating up to 2.9 times higher wall shear forces
compared to the plain empty tubes as seen in table 1.
Tab. 1: Shear force multiplier when using one particular
insert geometry
Reynolds du /dy [1/s] Multiplier [-]
100 0.4 2
500 0.55 2.75
3000 0.8 2.9
WaU shear stress calculations by measuring the forces
impacting on the insert
In a separate eχperimental setup the wall shear forces
were measured directly over the whole insert length.
r form drag
LL
・’
.’
・”
.・’
.・”
”
.”
・’
ふ
’
y・
‐
ぐ・l
″‐
に
’・
-
Fluid flow
Tube wall
JQ
むZ
ψ
=QS
↑
U。
Fig..6: Measurement principle to determine the increased
wall shear force created with the addition of
hiTRAN insert
In this setup the insert is installed in a tube in such a
way that the loops are just touching the tube wall, without
exerting any pressure. In order to measure a large range of
fluid conditions water and water-glycerin mixtures were
used. A load cell measures the form drag inflicted by the
fluid flow on the insert (fig. 6). In addition the pressure drop
over the installed insert length is measured. The overall wall
shear force can then calculated as shown in Equation (5).
^ wall shear =^P * ^cross section ‾F form drag (5)
The wall shear stress (Twaiu)is calculated by dividing the
measured shear force (Fwall shear)with the tube surface area
occupied by the insert. The dimensionless “wall" friction
factor for a particular insert geometry can then be calculated
as follows:
ル
一・I jo
p
e
i u
o
jp
u
i iiem
-
-
Ξ包
p u^
1
0.1
0.01
(6)
圖皿鞠~ こ二こ■■●nn・~
ぺ; ≪!!iiiこ
0.001
100 1000 10000 100000
Reynolds Number [-1
Fig.フ: Measured wall fanning friction factor for a
particular hiTRAN geometry compared with plain
empty tube data as fianction of Reynolds number
www.heatexchanger-fouling.com 374
Drogemij ‖er et al. /Increased shear, reduced wall temperatures, use of
Using this approach the dimensionless wall friction
factor for hiTRAN inserts can be measured and compared
with the plain empty friction factor as demonstrated in
figureフ。
By connecting the data points, correlation of the wall
friction factor for each insert geometry, can be found. In
figure 8 the increase in wall shear force compared to a plain
empty tube is shown and eχpressed as multiplier. The area
between the blue and red line represents the hiTRAN
operation area with low and high density inserts。
8
6
4
2
一-1 jaiidmni/
M
0
Fig. 8
10 100 1000 10000 100000
Reynolds Number [・】
Shear stress multiplier when using hiTRAN
inserts. Bottom line (Red)low dense insert. Top
line (blue)high dense insert, yellow points data
from LDV experiments
As seen in figure 8 the highest increase can be found
just before the transition from laminar to turbulent flow.
Here up to 7 times higher shear stress is measured by
comparison to the plain empty tube. The measured data for
a medium dense insert from the Laser eχperiments in table l
are also shown and indicate good agreement with the
results.
WaU shear stress simulations with CFD
Apart from the two experimental methods CFD was
employed to simulate the flow behaviour with inserts. The
CFD package ANSYS CFX was used to carry out the
simulations presented in this paper. The insert geometries
were drawn using ANSYS DesignModeler. Some
simplifications were made to the insert geometries to make
them easier to mesh. The geometries were then meshed
using the ANSYS meshing software. A mesh independence
study was carried out on each of the geometries to ensure
that the mesh was of high enough quality to accurately
model each of the inserts. The Shear Stress Transport (SST)
turbulence model was used for the turbulent regime, since it
gave more accurate near wall values compared to the k-E
turbulence model。
The calculations were performed under isothermal flow
conditions and with property, geometry and process
conditions similar to the shear force experiments. Figure 9
shows a typical simulated velocity profile in laminar flow,
which resembles very much the LDV measured flow profile
(fig 4), with steep velocity gradient near to the wall.
{s
/iu}
A
ip
o|a
A
1
-12 -8 -a 0 4 8 12
Radial Position 【mm]
Fig. 9: Velocity profile with CFD simulation, Reynolds
1000
To compare the simulated results with the measured
data the local wall shear stress values is integrated over the
whole tube surface area. The results then indicate the
average CFD simulated shear stress. In figure 10 the
measured correlated wall shear stress is compared for
different Reyno ↓ds numbers with the CFD simulated data
for discrete data points.
(Sl
)SS9J1S
」cans
IIBM
50 ●CFD high packing
0
0
0
0
0
4
3
2
1
density
●CFD medium packing
density //
』 ’
。 /*
一”””
♂ /
』
//
´ ◆
/♂/
●=.*
〆
5000 25000 45000
Reynolds Number 【-】
Fig. 10: Wall shear stress correlations from measured data
for medium dense (red dashed line)and high dense
(blue dashed line)hiTRAN inserts with CFD
simulations
For the dense inserts the CFD simulated values are in
very good agreement with the correlated measurement data.
CFD simulated shear stress values are slightly higher
compared to the measured data for the medium dense insert.
Having validated the CFD simulation against the measured
data, in a next step the local shear stress can be investigated.
In contrast to a plain empty tube the wall shear stress in a
hiTRAN enhanced tube varies along the tube wall. This is
shown in figure l l were for a medium dense insert the shear
stress is shown between two adjacent loops。
The shear stress peaks directly above the 100p wire and
is at its lowest just after the wire. This can be confirmed
when doing experiments with water particle suspension. In
figure 12 it is evident that some particles are settling out in
low shear areas behind the loops but principally in the plain
empty tube section.
www.heatexchanger-fouling.com 375
0
0
0
0
0
0
o
a) to
4
2
1?
a
)ssojis
Jcous
liewL
0
Heat Exchanger Fouling and Cleaning - 2013
=medium dense hiTRAN
2 4 6 8
Position between loops (mm)
Fig. 11: Shear stress distribution between adjacent hiTRAN
loops. Water, flow velocity 2.06m/sec Reynolds
number 30000
Fig. 12: Water flow suspension. Average particle size50Lim, density 2420kg/m3
The simulation also shows that the lowest shear stress
calculated just behind the 100p wire is of the same
magnitude as the plain empty tube value (red line in fig. 11)
CASE STUDY; VDU FEED EXCHANGER
The findings show that hiTRAN inserts influence tube
wall temperatures and tube wall shear rates and can be used
in order to change fouling behaviour。
In a retrofit application hiTRAN wire matrix elements
were installed in a two in series AES type Feed / Effluent
Eχchanger. In these eχchangers the tube side fluid ( VDU
feed)is heated. before entering the fired heater at about
300 °C. Under design conditions the exchanger operates in
the transitional flow regime of Re ~6000. The overall heat
transfer rate of the exchanger is tube side controlled. On the
sheU side Vacuum Residue is cooled from 350 °C, to preheat
the VDU feed。
The exchanger was suffering from very low tube side
design velocities of around OAm/s, resulting in low tube
side coefficients with high wall temperatures, long fluid
residence time and low wall shear stress. The low tube side
frictional pressure drop also increases the risk of fluid mal
distribution within the bundle.
10
Fig. 13: Investigation of fouled AES VDU / Feed heatexchanger.
Figure 13 shows the fouled plain empty tube eχchanger
after operation of 1130days. Fouling on the tube side and
the shell side was observed. Before installing a hiTRAN
Thermal System the exchanger was cleaned. After installing
hiTRAN the calculated tube side pressure drop increased
within the allowable limit to about 70mbar.. The induced
wall shear increase from 0.43Pa to about 0.75Pa. More
importantly the tube side heat transfer rate increased by
about four times from 3 12W/m^K to 1 132W/m^K. As a
result, the main heat transfer resistance shifts to the shell
side. Process information with hiTRAN installed, was taken
over a period of 10 month. In order to compare both setups
only the initial period of 300 days were used for evaluation.
The initial setup was tube side controlled and was in an area
where hiTRAN Systems can improve the heat transfer
substantially. The measured overall coefficient for the
exchangers was increased by about 40% (fig. 14).
0000000000
505050505
44332211
【y\z
^iMi o
iエ○
Udirty hiTRAN
=average hiTRAN 刈00 days
Udirty pl3in empty
|
|
| ・ 。l j
^i1 'MMJ・L.A.a It . | , IT
| 。k T L71711 r7
rr^'「 日 仇1ー・y nit'・ 1,ノ 1^' r ・ r VV ●7
" VT ・ ・ ・ l ’
0
100
time [day5]
200 300
Fig. 14: Overall heat transfer coefficient over time for plain
empty tube and hiTRAN System
www.heatexchanger-fouling.com 376
Drogemij ‖er et al. /Increased shear, reduced wall temperatures, use of
Since the tube side heat transfer was increased there
was also an eχpected reduction in tube side wall
temperature.
0
0
0
0
0
0
0
0
0
0
4
3 2
1
0
9
8
7
6
5
3
3
3
3
3
2
2
2
2
2
r
】Q」コleisdE
31 liewi
,砥,i a. I. iI
LML 直 岨 為匿 ゴ レ__
'AU. l。,,4 1. . X
り 胴ijJ、;..LJ^a.^^
71 ■ 皿1-1 軍●・ .ー 、¶ Tl ゛I 皿7 W ’fr 専‾
1
|-
i/TU u 享゙ \八J n VI
' nf ¶ ゛Ij
μ1 「 -plain empty
-hJTRAN
0 100 200 300time 【days 】
Fig.l5: Tube wall temperatures before and after
installation of hiTRAN
The average wall temperature was about 13 °C lower
compared to the plain empty tube measurements. Even
greater was the reduction in peak values from 330 °C to
310 °C(fig. 15)・
Tab. 2: Conditions before and after installing hiTRAN
Geometry Plain
empty
hiTRAN
TEMA Type AES
Shells in Series 2
Tube geometry 1416 × 19.05mm ODx6096mm
No of tube passes 2
Process
Averages over time [days] 1130 300
Shell side flow [kg/s] 28.6 25.6
Tube side flow [kg/s] 33.4 30
Shell side in/out Temp [゜c] 347 /293 343 /286
Tube side in/out Temp[ ゜c] 256 /294 268 /309
Duty[MW] 3.79 3.54
EMTD[゜c] 41.2 18.3
Average tube wall temp. [゜c] 309 296
Tube side HTC [W/m2K] 312 1132
Shell side HTC [W/m2K] 647 602
Tube side velocity [㎡s] 0.34 0.31
Overall coefficient [W/m2K] 147 206
Tube side Reynolds [-] 4842 5018
dp calculated [bar] 0.09 0.7
Wall shear[Pa] 0.43 0.75
Over a period of 300 days the fouling factor was
calculated and compared to the longer runtime prior to the
installation of hiTRAN inserts. In figure 16 it can be seen
that the initial fouling is similar to the plain empty tube
behaviour. It will be interesting to see whether a jump in
fouling as experienced after about two years, can be avoided
with the installed hiTRAN Elements. For the runtime so far
the benefit can be seen in the increased duty of the
exchanger, reducing the load on the fired heater.
[M
P
1;
U
J] jo
p
e
j
6
ui|n
o
j
0.02
0.015
0.01
0.005
0
-0.005
-0,01
-0,015
-0.02
time 【days 】
Fig. 16: Fouling factor for plain empty tube and hiTRAN
operation
CONCLUSION
This paper discusses the possibility of altering the plain
empty tube conditions concerning wall temperature and
wall shear stress by using hiTRAN tube inserts. For fouling
threshold models, the knowledge of these two variables is
necessary in order to determine the eχpected fouling rate。
It is shown that the wall shear stress can be increased by
up to 7 times depending on the flow regime. Different
experimental set ups to measure wall shear forces directly
are presented and show good agreement. The measured
results give similar results to the CFD simulations. Those
simulations are ideal to investigate the local shear stress
distribution between the insert loops; again a qualitative
agreement with experimental results in particulate fouling
was found。
The impact on eχchanger performance is demonstrated
in a heat exchanger revamp scenario. The overall operation
time evaluated with hiTRAN is too short in order to draw
definite conclusion on the impact of the fouling behaviour
with enhancement installed. Nevertheless it is clear that the
process benefits from the increased overall heat transfer。
For turbulent flow hiTRAN Systems can be a viable
option in cases where the exchanger operates at low flow
velocities and does not utilize all available pressure drop,
for eχample using a single pass arrangements to achieve
small driving temperature differences. Sufficient research
has been completed to provide confidence in applying
hiTRAN Systems in many fouling duties. More research
will be needed in order to verify whether there are
additional effects to consider when modeling the fouling
rate behaviour in the presence of tube internals.
www.heatexchanger-fouling.com 377
NOMENCLATURE
NunApPrR
Rt
TfT
U
y
Heat Exchanger Fouling and Cleaning - 2013
Area,m^
diameter, m
fouling model activation energy, J/mol
drag force, N
shear force, N
heat transfer coefficient, W/m^K
length, m
Nusselt, dimensionless number
Prandtl eχponent
pressure drop, N/m^
Prandtl, dimensionless number
gas constant, J/mol K
fouling resistance, m^KAV
time, s
film temperature, K
temperature, K
velocity, m/s
coordinate, m
Greek letters
αnp
T
ヂ
T
a
p
fouling model parameter, m^K/J
fouling model parameter, dimensionless
fouling model parameter, m^K/J Pa
fanning friction factor, dimensionless
shear stress, Pa
dynamic viscosity, Pa s
density, kg/m^
Subscript
1 inner
o outer
w wall
REFERENCES
Ebert, W., and Panchal, C.B., 1997, Analysis of Exxon
crude oil slip stream coking data. Proc. Mitigation of
Fouling in Industrial Heat Eχchanger Equipment, San Luis
Obispo, USA Begell house pp.45 1 - 460
Gough, M.J., I.J. Gibbard, G.T. Polley and A.S.
McMuUan, 1995, Case studies of refinery fouling
reduction. Proc. Engineering Foundation Conference on
Fouling Mitigation of Industrial Heat Eχchangers.
California.
Meneyan, Y., Crittenden, B. 2012, Fouling thresholds
in bare tubes and tubes fitted with inserts, Applied Energy,
Volume 89(1), pp. 67 -73
Ritchie, J.M 。Droegemueller, P., Simmons M.J.H.,
2007, hiTRAN Wire Matrix Inserts in Fouling Applications
in Heat Exchanger Fouling and Cleaning VII, Eds, ECI
Symposium Series, Volume RP5.
Smeethe, A., P. Droegemueller, J. Wood, W. Bujalski,
2004, Fluid dynamics in a tube equipped with wire matrix
inserts. Proc. 4th European Thermal Sciences Conference.
Birmingham
Wilson, D.I., PoUey, G.T., Pugh, S.J.,
2005, Proceedings of the 6'* International Conference on
Heat Exchanger Fouling and Cleaning- ECI Symposium
Series, Kolster Irsee, Germany, June 5 - 10, Volume RP2
www.heatexchanger-fouling.com 378